In [1]:
import math 
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
sns.set()
In [2]:
dataset = pd.read_csv('./cleaned_df_with_manufact.csv', keep_default_na=False) #Avoid 'null' in CSV to be parsed as NaN 
dataset.describe()
Out[2]:
price year odometer
count 83003.000000 83003.000000 83003.000000
mean 13516.885438 2011.128610 103821.619303
std 8970.604066 5.800385 61473.967556
min 1050.000000 1981.000000 0.000000
25% 6499.000000 2007.000000 54000.000000
50% 11500.000000 2012.000000 99748.000000
75% 18448.000000 2016.000000 144200.000000
max 40712.000000 2021.000000 410374.000000
In [3]:
dataset.head(3)
Out[3]:
price year manufacturer model condition cylinders fuel odometer transmission drive type paint_color state
0 13995 2012.0 ford f-150 good 6 cylinders gas 188406.0 automatic 4wd truck grey mn
1 7995 2010.0 chevrolet equinox good 4 cylinders gas 108124.0 automatic 4wd SUV grey mn
2 8995 2011.0 chevrolet traverse good 6 cylinders gas 178054.0 automatic 4wd SUV white mn
In [4]:
dataset.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 83003 entries, 0 to 83002
Data columns (total 13 columns):
 #   Column        Non-Null Count  Dtype  
---  ------        --------------  -----  
 0   price         83003 non-null  int64  
 1   year          83003 non-null  float64
 2   manufacturer  83003 non-null  object 
 3   model         83003 non-null  object 
 4   condition     83003 non-null  object 
 5   cylinders     83003 non-null  object 
 6   fuel          83003 non-null  object 
 7   odometer      83003 non-null  float64
 8   transmission  83003 non-null  object 
 9   drive         83003 non-null  object 
 10  type          83003 non-null  object 
 11  paint_color   83003 non-null  object 
 12  state         83003 non-null  object 
dtypes: float64(2), int64(1), object(10)
memory usage: 8.2+ MB

Individual Variable Description

Price

In [5]:
fig, ax = plt.subplots(figsize=(8,6))
ax.hist(dataset['price'], 100)
ax.set_title('Distribution of Price'), ax.set_xlabel('Price'), ax.set_ylabel('Count')
fig.show()

Year

In [6]:
dataset['year'].mean()
Out[6]:
2011.128609809284
In [7]:
fig, ax = plt.subplots(figsize=(8,6))
ax.hist(dataset['year'], len(dataset['year'].value_counts()), align='right')
ax.set_title('Distribution of Vehicle Year'), ax.set_xlabel('Year'), ax.set_ylabel('Count')
ax.arrow(2009.2,5900,0,-2500,head_width=0.5,head_length=150,color='gray')
ax.axvline(x=dataset['year'].mean(),color='black',label='mean')
ax.axvline(x=dataset['year'].median(),color='r',label='median')
ax.text(2017,8300,'2017',horizontalalignment='center')
ax.text(2009,6000,'2009',horizontalalignment='center')
ax.legend()
fig.show()

Make

In [8]:
fig, ax = plt.subplots(figsize=(8,6))
ax.barh(dataset['manufacturer'].value_counts().iloc[:10].index, dataset['model'].value_counts().iloc[:10].values)
ax.set_title('Top 10 Popular Make'), ax.set_xlabel('Count'), ax.invert_yaxis()
fig.show()

Model

In [8]:
fig, ax = plt.subplots(figsize=(8,6))
ax.barh(dataset['model'].value_counts().iloc[:10].index, dataset['model'].value_counts().iloc[:10].values)
ax.set_title('Top 10 Popular Model'), ax.set_xlabel('Count'), ax.invert_yaxis()
fig.show()

Condition

In [9]:
fig, ax = plt.subplots(figsize=(8,6))
ax.barh(dataset['condition'].value_counts().index, dataset['condition'].value_counts().values)
ax.set_title('Vehicle Condition'), ax.set_xlabel('Count'), ax.invert_yaxis()
fig.show()

Clearly, a large amount (25416) of vehicle condition are not reported in this dataset.

Engine Cylinder Count

In [10]:
fig, ax = plt.subplots(figsize=(8,6))
ax.barh(dataset['cylinders'].value_counts().index, dataset['cylinders'].value_counts().values)
ax.set_title('Engine Cylinder Count'), ax.set_xlabel('Count'), ax.invert_yaxis()
fig.show()

Fuel Type

In [11]:
fig, ax = plt.subplots(figsize=(8,6))
ax.bar(dataset['fuel'].value_counts().index, dataset['fuel'].value_counts().values)
ax.set_title('Fuel Type'), ax.set_ylabel('Count')
fig.show()

Mileage

In [12]:
fig, ax = plt.subplots(figsize=(8,6))
ax.hist(dataset['odometer'], 100, align='right')
ax.set_title('Mileage'), ax.set_xlabel('Mile'), ax.set_ylabel('Count')
ax.axvline(x=dataset['odometer'].mean(),color='black',label='mean')
ax.axvline(x=dataset['odometer'].median(),color='r',label='median')
ax.legend()
fig.show()

Transmission

In [13]:
fig, ax = plt.subplots(figsize=(8,6))
ax.bar(dataset['transmission'].value_counts().index, dataset['transmission'].value_counts().values)
ax.set_title('Transmission Type'), ax.set_ylabel('Count')
fig.show()

Drive

In [14]:
fig, ax = plt.subplots(figsize=(8,6))
ax.bar(dataset['drive'].value_counts().index, dataset['drive'].value_counts().values)
ax.set_title('Drive Type'), ax.set_ylabel('Count')
fig.show()

Vehicle Type

In [15]:
fig, ax = plt.subplots(figsize=(8,6))
ax.barh(dataset['type'].value_counts().index, dataset['type'].value_counts().values)
ax.set_title('Vehicle Type'), ax.set_xlabel('Count'), ax.invert_yaxis()
fig.show()

Paint Color

In [16]:
fig, ax = plt.subplots(figsize=(8,6))
ax.barh(dataset['paint_color'].value_counts().index, dataset['paint_color'].value_counts().values)
ax.set_title('Paint Color'), ax.set_xlabel('Count'), ax.invert_yaxis()
fig.show()

State

In [17]:
fig, ax = plt.subplots(figsize=(8,8))
ax.barh(dataset['state'].value_counts().index, dataset['state'].value_counts().values)
ax.set_title('State'), ax.set_ylabel('Count'), ax.invert_yaxis()
fig.tight_layout()
fig.show()

Multivariate Analysis

In [18]:
sns.pairplot(pd.concat([dataset.price,dataset.odometer,dataset.year], axis=1),              corner=True, plot_kws=dict(marker="+", linewidth=1))
plt.show()
In [19]:
with sns.axes_style('white'):
    sns.jointplot('price', 'odometer', dataset, kind='hex')
In [20]:
with sns.axes_style('white'):
    sns.jointplot('price', 'year', dataset, kind='hex')
In [21]:
#sns.regplot('odometer','price',dataset)
In [22]:
dd = pd.concat([dataset,pd.get_dummies(dataset.condition)], axis=1)
dd = dd.drop(columns=['null'])
dd = pd.concat([dd,pd.get_dummies(dataset.cylinders)], axis=1)
dd = dd.drop(columns=['null'])
dd = pd.concat([dd,pd.get_dummies(dataset.fuel)], axis=1)
dd = pd.concat([dd,pd.get_dummies(dataset.transmission)], axis=1)
dd = pd.concat([dd,pd.get_dummies(dataset.drive)], axis=1)
dd = pd.concat([dd,pd.get_dummies(dataset.type)], axis=1)
#dd = pd.concat([dd,pd.get_dummies(dataset.paint_color)], axis=1)

corr = dd.corr()
fig, ax = plt.subplots(figsize=(11, 9))
sns.heatmap(corr, mask=np.triu(np.ones_like(corr, dtype=bool)), center=0,
            square=True, linewidths=.1, cbar_kws={"shrink": .5})
fig.show()
In [23]:
corr = dataset.corr()
fig, ax = plt.subplots(figsize=(11, 9))
sns.heatmap(corr, center=0, annot=True,
            square=True, linewidths=.1, cbar_kws={"shrink": .5})
fig.show()
In [24]:
dataset.head(3)
Out[24]:
price year model condition cylinders fuel odometer transmission drive type paint_color state
0 13995 2012.0 f-150 good 6 cylinders gas 188406.0 automatic 4wd truck grey mn
1 7995 2010.0 equinox good 4 cylinders gas 108124.0 automatic 4wd SUV grey mn
2 8995 2011.0 traverse good 6 cylinders gas 178054.0 automatic 4wd SUV white mn

Price and Odometer vs. Condition

In [25]:
sns.catplot(x='price',y='condition',data=dataset[['price','condition']],                   kind='boxen')
plt.show()
In [26]:
sns.catplot(x='odometer',y='condition',data=dataset[['odometer','condition']],             kind='violin')
plt.show()

Price vs. Vehicle Type

In [27]:
sns.catplot(x='price',y='type',data=dataset[['price','type']],                             kind='boxen')
plt.show()

Price vs. Transmission, Drive, Cylinders and Fuel Type

In [28]:
sns.catplot(x='price',y='fuel',col='drive',hue='transmission',
            data=dataset[['price','transmission','drive','fuel']],kind='boxen')
plt.show()
In [29]:
sns.catplot(x='price',y='cylinders',data=dataset[['price','cylinders']],               kind='boxen')
plt.show()

Price vs. Paint Color

In [30]:
sns.catplot(x='price',y='paint_color',data=dataset[['price','paint_color']],               kind='boxen',palette=sns.color_palette(['gray','white','lime','silver','blue','brown','purple','red','black','orange','green','yellow']))
plt.show()
In [ ]: